Improved estimators for simultaneous estimation of variance components

被引：0

作者：

Klonecki, W ^{[1
]}

Zontek, S ^{[1
]}

机构：

[1] POLISH ACAD SCI,INST MATH,PL-51617 WROCLAW,POLAND

来源：

STATISTICS & PROBABILITY LETTERS | 1996年 / 29卷 / 01期

关键词：

variance components; quadratic risk function; improved estimators; Bayesian estimators;

D O I：

10.1016/0167-7152(95)00153-0

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

Simultaneous estimation of variance components under quadratic risk is discussed. The estimators considered are scale preserving and location invariant and permit a simple closed expression for risk. For several one-way random normal models, nonnegative estimators are constructed which are comparable to the maximum likelihood estimators and to some of Portnoy's (1971) estimators.

引用

页码：33 / 43

页数：11

共 12 条

[1]

Box GE., 2011, BAYESIAN INFERENCE S

[2]

BROMELING LD, 1985, BAYESIAN ANAL LINEAR

[3]

DASGUPTA A, 1986, ANN STAT, V14, P206

[4] ALL ADMISSIBLE LINEAR ESTIMATORS OF THE VECTOR OF GAMMA-SCALE PARAMETERS WITH APPLICATION TO RANDOM EFFECTS MODELS [J].